Immune checkpoint inhibitors for glioblastoma: emerging science, clinical advances, and future directions

Abstract

Glioblastoma (GBM), the most common and aggressive primary central nervous system (CNS) tumor in adults, continues to have a dismal prognosis. Across hundreds of clinical trials, few novel approaches have translated to clinical practice while survival has improved by only a few months over the past three decades. Randomized controlled trials of immune checkpoint inhibitors (ICIs), which have seen impressive success for advanced or metastatic extracranial solid tumors, have so far failed to demonstrate a clinical benefit for patients with GBM. This has been secondary to GBM heterogeneity, the unique immunosuppressive CNS microenvironment, immune-evasive strategies by cancer cells, and the rapid evolution of tumor on therapy. This review aims to summarize findings from major clinical trials of ICIs for GBM, review historic failures, and describe currently promising avenues of investigation. We explore the biological mechanisms driving ICI responses, focusing on the role of the tumor microenvironment, immune evasion, and molecular biomarkers. Beyond conventional monotherapy approaches targeting PD-1, PD-L1, CTLA-4, we describe emerging approaches for GBM, such as dual-agent ICIs, and combination of ICIs with oncolytic virotherapy, antigenic peptide vaccines, chimeric antigenic receptor (CAR) T-cell therapy, along with nanoparticle-based delivery systems to enhance ICI efficacy. We highlight potential strategies for improving patient selection and treatment personalization, along with real-time, longitudinal monitoring of therapeutic responses through advanced imaging and liquid biopsy techniques. Integrated radiomics, tissue, and plasma-based analyses, may potentially uncover immunotherapeutic response signatures, enabling early, adaptive therapeutic adjustments. By specifically targeting current therapeutic challenges, outcomes for GBM patients may potentially be improved.

Introduction

Glioblastoma (GBM), the most common and aggressive primary central nervous system (CNS) tumor in adults, continues to have a dismal prognosis and remains inevitably fatal [1, 2]. Despite the thousands of patients enrolled in GBM trials [3] few new approaches have translated to clinical practice and survival has improved by only a few months over the past three decades [4].

The utility of novel therapies for GBM remains limited secondary to the blood-brain barrier (BBB) limiting sufficient intracranial drug delivery [3, 5], the unique tumor microenvironment (TME) [6], tumor heterogeneity [7, 8], the evolution of tumor molecular and TME characteristics on therapy [9], and the challenges in obtaining longitudinal information (Fig. 1) [10].

Fig. 1.

Challenges to effective therapies for patients with glioblastoma. “The figure illustrates the distinctive characteristics of GBM (WHO grade 4) that are understood to hinder the development of effective anti-tumor therapies. These include (1) an anatomical location shielded by the blood–brain barrier, (2) intra- and inter-patient tumor heterogeneity, (3) low tumor mutation burden (4) infiltrative behavior, and (5) a highly immunosuppressive TME. The latter showcases the presence of GBM-driven cytokines with immunosuppressive and tumor-promoting properties, along with immunosuppressive cell populations such as Tregs and M1-like TAMs, accompanied by upregulated exhaustion markers. Additionally, GBMs strategically downregulate antigen-processing and presentation molecules to effectively evade T cell activation. Abbreviations APC, antigen-presenting cell; bFGF, basic fibroblast growth factor; ECM, extracellular matrix; GBM, glioblastoma; IL, interleukin; LAG-3, lymphocyte-activation gene 3; MDSC, myeloid-derived suppressor cells; PD-1, programmed cell death protein 1; PGE2, prostaglandin E2; TAM, tumor-associated microglia and macrophages; TGF-β, transforming growth factor-β; TIM-3, T-cell immunoglobulin and mucin domain; TME, tumor microenvironment; Treg, regulatory T cell; WHO, the World Health Organization.” Reproduced under Creative Commons Attribution 4.0 International License from Salvato and Marchini, 2024

Trials of immune checkpoint inhibitors (ICIs), which have led to impressive clinical success for other advanced or metastatic solid tumors, have so far failed to demonstrate a clinical benefit for GBM. Given the historical failures and the need for careful investigation, this review aims to summarize lessons from major clinical trials of ICIs for GBM and describe currently promising avenues of investigation. We explore the biological mechanisms driving ICI responses, focusing on the role of TME, immune evasion, and molecular biomarkers. We provide an overview of the current science through pivotal clinical trials, discuss ongoing research and novel therapeutic strategies, examine new immune checkpoint targets and resistance mechanisms, explore the role of liquid biopsy and optimal sequencing of ICIs. Beyond conventional monotherapy approaches targeting PD-1, PD-L1, CTLA-4, we also discuss innovative approaches and sequencing of ICIs. Finally, we outline strategies for overcoming current therapeutic challenges and optimizing future clinical trials.

State of the evidence

The clinical outcomes of ICIs for GBM patients have been variable and often disappointing, while our understanding of the optimal sequence of immunotherapy in the setting of other therapeutic interventions has also been limited.

Single agent ICIs

In the newly diagnosed setting, CheckMate-498 and CheckMate-548 have been the largest phase 3 RCTs to date. CheckMate-498 (NCT02617589) evaluated nivolumab + radiation versus temozolomide + radiation in MGMT promoter unmethylated GBM. The nivolumab arm was inferior with patients experiencing lower median overall survival (mOS) and higher frequencies of treatment related adverse events (TRAE) [11]. Some concerns were also raised regarding the insufficient accuracy of MGMT testing and consequent impact on adequate patient selection, hindering downstream response. CheckMate-548 (NCT02667587) investigated the utility of adding nivolumab to radiation and temozolomide in newly diagnosed GBM patients with a methylated or indeterminate MGMT promoter status. No benefit in progression free survival (PFS) and OS and higher TRAEs were noted in the nivolumab cohort [12].

In the recurrent setting, two Phase II RCTs have investigated the combined benefit of pembrolizumab with bevacizumab, showing a PFS but no OS benefit, similar to other bevacizumab clinical trials (NCT02017717, NCT02337491) [13, 14]. CheckMate-143 (NCT02017717) was a Phase III RCT evaluating nivolumab against bevacizumab in the same patient population, showing similar median survival, longer PFS with bevacizumab, and overall similar safety profiles [15]. A recent phase II trial (NAVAL, NCT03452579) compared nivolumab combined with either standard-dose or low-dose bevacizumab for recurrent GBM. The trial demonstrated similar PFS and OS between the two arms, with post-hoc analysis showing a survival benefit for patients over 60 years receiving standard-dose bevacizumab. Notably, the standard-dose arm also exhibited increased systemic inflammatory responses and reduced immunosuppressive MDSCs, suggesting potential differences in immune modulation across the dosing strategies. Neoadjuvant anti-PD-1 immunotherapy led to a significant increase in systemic inflammatory responses, marked by elevated levels of proinflammatory cytokines such as IFN-γ and TNF-α, which enhanced T-cell activation. Moreover, there was a notable reduction in MDSCs, which are known to inhibit T-cell responses through mechanisms like arginase-1 and iNOS expression. This reduction in MDSCs further amplified the immune system’s capacity to target and destroy tumor cells [16].

Neoadjuvant single-agent ICI

Neoadjuvant administration of ICI, in patients who are candidates for tumor resection, may enhance anti-tumor response due to surgically-induced cellular disruption and increased presentation of tumor antigens in the setting of an activated immune system, in addition to increased inflammatory activity at resection site. In a single-arm trial of newly diagnosed (n = 3) and recurrent GBM (n = 27), neoadjuvant pembrolizumab led to significant immune activation (NCT02550249); [17] while OS in 2 of 3 newly diagnosed GBM was also extended. Similarly, the multi-institutional trial by the Ivy Foundation Early Phase Clinical Trials Consortium (IFEPCTC) evaluated neoadjuvant pembrolizumab in 35 recurrent GBM patients, revealing enhanced local and systemic immune responses [15]. The study found that in these 35 patients, neoadjuvant anti-PD-1 therapy (pembrolizumab) led to significant immune activation, both locally and systemically. The local immune response was characterized by increased infiltration of CD8 + T cells and MHC class II upregulation within the TME. Systemically, patients showed elevated levels of IFN-γ and IL-2, indicating a strong systemic immune response. This dual immune activation was associated with improved survival outcomes.

However, the application of neoadjuvant ICI therapy presents several challenges that must be carefully managed, including exacerbation of brain edema, which can result in sudden neurological decline, and delay or even preclude surgical resection [18]. Furthermore, determining the optimal number of ICI cycles before surgery is another significant challenge. While some studies suggest that 2–3 cycles might suffice to prime the immune system without inducing excessive toxicity, there is no consensus on the optimal regimen. Administering ICIs before surgery can add complexity to the overall treatment regimen and increase the burden on patients and families, particularly in those with symptomatic or rapidly progressing disease and significant comorbidities. Considering that most newly diagnosed GBM patients present following an acute neurological event, coordinating this may indeed be quite challenging [19]. Extending ICI treatment could also raise the risk of severe immune-related adverse events without providing additional therapeutic benefits [20]. Neoadjuvant ICI could logistically be a more viable option in the recurrent setting but the overall health of the patient and other treatment-mediated factors influencing tumor responses – as discussed below – are challenges that must be considered.

Dual-agent ICIs

The concurrent upregulation of multiple immune checkpoint molecules represents a possible resistance mechanism to PD-1 and other single agent ICI therapy in GBM. This has prompted exploration of rational combinations of PD-1 inhibitors with other checkpoint inhibitors. In one such combination, a phase 1 trial focusing on post operative patients with recurrent GBM combined a PD-1 inhibitor (nivolumab) with anti-LAG3 (NCT02658981) [21], demonstrating tolerability and early evidence supporting the concept that targeting multiple checkpoints might help overcome resistance to single-agent therapies. Another trial, (NCT04145115) examined the combination of nivolumab and ipilimumab in patients with recurrent GBM [22]. This Phase II trial explored the potential synergistic effects of these two ICIs, which has been a paradigm changing approach in melanoma [23].

Understanding the complex interplay between various immune evasion and/or suppression mechanisms at play within and outside the TME is crucial for developing more effective and tailored immunotherapeutic strategies. Enhanced understanding of these mechanisms is crucial for optimizing treatment and monitoring protocols [24, 25].

Role of the tumor microenvironment

The immunosuppressive TME of GBM includes regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs), and tumor-associated macrophages (TAMs), which collectively inhibit effective anti-tumor immune responses (Figs. 2 and 3) [26]. Tumor cells also express immune checkpoints, such as PD-L1, CTLA-4, and other T cell exhaustion markers [25]. Temozolomide has been shown to increase the proportion of exhausted T cells in GBM, thereby reducing the efficacy of checkpoint inhibition in pre-clinical models [27]. Thus, modulating the temozolomide dose could differentially affect T-cell responses.

Fig. 2.

Dynamic crosstalk between tumor and immune cells as a potential contributor to an enhanced response to immune checkpoint blockade “Different transcriptional programs are defined by the abundance of immune cells such as macrophages, microglia, and T cells as well as the immunogenicity of tumor cells that includes the expression of MHC class I and II. In addition, the generation of a peripheral immune response following immune checkpoint inhibitors is a critical component of a successful therapeutic response”. Reproduced under Creative Commons Attribution 4.0 International License from Arrieta et al. 2023

Fig. 3.

Summary of the potential mechanisms that contribute to the response to ICIs in patients with glioblastoma. “Three potential mechanisms have been associated with better survival in patients with GBM treated with anti–PD-1 therapy: activation of the ERK1/2 cascade of the MAPK pathway; defects in the replication stress response in tumor cells; and germline POLE mutations. In addition, the mechanism underlying intratumoral Treg depletion is shown in the context of anti–CTLA-4 immunotherapy induced by antigen-dependent, cell-mediated cytotoxicity, which has been associated with germline polymorphisms of Fc-γR with high binding affinity to the therapeutic monoclonal antibodies. AP-1, activator protein 1; A-RAF, A-rapidly accelerated fibrosarcoma; ATM, ataxia telangiectasia mutated; ATR, ataxia telangiectasia and Rad3-related protein; CHK1, checkpoint kinase 1; FOSL1, Fos-related antigen 1; GRB2, growth factor receptor–bound protein 2; NF1, neurofibromin 1; SHP-2, Src homology region 2 domain–containing phosphatase-2; SOS, Son of Sevenless.” Reproduced under Creative Commons Attribution 4.0 International License from Arrieta et al. 2023

Infiltration of activated T cells into the TME is critical for the success of immunotherapy. The IFEPCTC multicenter RCT demonstrated that neoadjuvant anti-PD-1 therapy did induce robust intratumoral and systemic immune responses, including elevated IFN-γ levels and increased CD8 + T cell infiltration [15]. This corroborated earlier animal studies, showing that anti-PD-1 blockade combined with stereotactic radiosurgery (SRS) significantly increased the CD8 + T cell-to-Treg ratio, correlating with improved survival outcomes in glioma models [28]. However, these results did not translate in a window-of-opportunity Phase 2 clinical trial (NCT02337686), wherein preoperative administration of pembrolizumab in recurrent GBM was associated with increased tumor-supporting macrophages, which do contribute to resistance against ICIs. These macrophage clusters expressed immunosuppressive markers, including PD-L1, VISTA, and B7H3, as determined by single-cell CyTOF analysis, which likely contributed to resistance against ICIs [29]. These discordant findings highlight the complexity of faithfully reproducing all characteristics of human glioma-immune interactions in animal models. They also reflect variability in immune responses among GBM patients and underscore the need to identify patient-specific predictive signatures.

Targeting the “cold” TME, characterized by low immune cell infiltration, high levels of immunosuppressive cells, and limited antigen presentation, is a critical strategy for overcoming resistance to ICI therapy in GBM [30]. However, the low tumor mutational burden (TMB) of GBM, combined with its extensive cellular and molecular heterogeneity, hinders anti-tumor immune responses [31]. Several promising approaches are being explored to modulate this hostile environment. One such approach focuses on glioma cells’ production of indoleamine 2,3-dioxygenase (IDO), an enzyme that depletes tryptophan, leading to the accumulation of Tregs and the inhibition of effector T-cell activity [23]. To counteract this, ongoing trials are combining PD-1 inhibitors with therapeutic cancer vaccines (NCT04013672, NCT03750071) to increase the infiltration of tumor-specific T cells and boost anti-tumor activity [32, 33],. A Phase 2a trial (N = 64) of SurVaxM, a peptide vaccine conjugate that activates anti-tumor immunity against survivin (highly expressed by GBM), has demonstrated utility in combination with TMZ for newly-diagnosed setting [32]. Final results from NCT04013672 combining pembrolizumab with SurVaxM for GBM patients at first recurrence are pending. Similarly, combining ICI with chimeric antigen receptor T-cell (CAR T) therapy and oncolytic viral therapy can also enhance the anti-tumor response by increasing immune cell infiltration and activation within the tumor [34, 35]. Combining ICIs like pembrolizumab with CAR T-cell therapy targeting EGFRvIII and oncolytic viral therapy, such as PVSRIPO, enhances the anti-tumor response by promoting increased immune cell infiltration, including CD8 + T cells, and activation within the TME, though early trials have shown challenges in sustaining long-term efficacy. While examining the details of these alternative immunotherapy approaches are beyond the scope of this review, CAR-T cells specifically target tumor antigens, boosting T cell presence in the TME, while oncolytic viruses selectively infect tumor cells and release antigens, stimulating a broader immune response. These approaches collectively work to convert the cold TME into one that is more responsive to immunotherapy. Additionally, trials are exploring combinations with specific targeted therapies such as the histone deacetylase inhibitor vorinostat, which enhances tumor immunogenicity by increasing the expression of MHC class I molecules and tumor-associated antigens, promoting CD8 + T cell infiltration and reducing immunosuppressive regulatory T cells in glioblastoma [36, 37]. The AKT inhibitor ipatasertib restores tumor immune recognition by inhibiting the PI3K/AKT pathway, downregulating PD-L1 expression, and reducing immunosuppressive cells such as MDSCs and Tregs, thereby synergizing with ICI therapy to boost antitumor immune responses.

Recent findings from multi-omics profiling of GBM TME highlight the significant role of fibrosis in TME post-treatment [38]. Here, multiple anti-glioma therapies, including anti-CSF-1R therapy, were found to preclinically and clinically induce a fibrotic response within the GBM TME, which paradoxically promoted tumor recurrence. The formation of fibrotic niches, mediated by fibroblast-like cells via TGF-β signaling post-treatment, creates a sanctuary for dormant tumor cells, shielding them from immune surveillance and aiding in long-term survival. Inhibition of this fibrosis, combined with anti-CSF-1R therapy, improved therapeutic outcomes in preclinical models [38]. These findings add another layer of complexity to TME’s role in therapeutic resistance, emphasizing the need for combination therapies that tackle not only immune suppression but also the structural modifications within the tumor bed that protect malignant cells.

Another promising immunotherapeutic strategy involves the modulation of polyamine metabolism, a pathway crucial for the survival of immunosuppressive myeloid cells within the TME [39]. Disrupting polyamine metabolism has been demonstrated to alter the immunosuppressive landscape, thereby enhancing the efficacy of ICIs [40]. Inhibiting polyamine biosynthesis using difluoromethylornithine (DFMO), a potent immune modulator, reduces the proliferation of immunosuppressive cells and synergizes with PD-1 blockade to significantly enhance anti-tumor immune responses in GBM models [41].

A recent advance in our understanding of the immune system’s interaction with the CNS and GBM involves the discovery that cranial bone marrow (CBM) in proximity to GBM tumors contains active lymphoid populations that significantly influence the anti-tumor immune response [42]. The presence of tumor-reactive CD8 + T cells within the CBM not only underscores the potential for immune surveillance within the CNS but also positions the cranioencephalic lymphoid units as crucial components of the TME [43]. Enhancing the immunogenic potential of CBM could potentially lead to more effective immunotherapeutic strategies.

Patient selection and molecular signatures of ICI response

While baseline molecular evaluation has now become the standard in non-resource-limited settings, molecular features of therapeutic response have not yet been well-characterized and validated [44]. Specifically, classic markers of ICI response seen in other cancers have yielded inconsistent results in GBM [45, 46].

The genetic diversity of solid tumors with mismatch repair (MMR) deficiency, and resultant mutational load, has been recognized to influence the response to anti-PD-1 immunotherapy [47]. Microsatellite instability (MSI) is a strong predictive biomarker for ICI response in other cancers, including colorectal and endometrial cancer. With rare exceptions, such as Lynch syndrome, MSI frequency is typically low in gliomas and most newly diagnosed GBM patients have a low mutational burden [48]. However, subsets of GBM patients exhibit a high TMB [49]. Mutations in the RAS/MAPK (ERK) pathway have been identified in replication repair-deficient hypermutated tumors, which confer sensitivity to MEK inhibition, while germline-driven replication repair-deficient high-grade gliomas exhibit unique hypomethylation patterns, further distinguishing these tumors from other GBM subsets [50, 51]. These hypermutated gliomas typically arise either due to constitutional defects in DNA polymerase and MMR genes or, more commonly, following temozolomide treatment in association with MMR defects. Hypermutation is reported in approximately 25% of temozolomide-treated gliomas and occurs more frequently in IDH mutant tumors. However, retrospective reviews indicate that hypermutation may not always confer a benefit from ICI therapy [49, 52, 53]. Work by the International Replication-Repair-Deficient (RRD) Consortium has recently demonstrated that amongst RRD high-grade glioma patients who fail anti-PD-1 monotherapy (N = 75), combination of immune-directed/synergistic salvage therapies led to improved outcomes (Fig. 4).

Fig. 4.

Emerging approaches and new frontiers in immune checkpoint inhibitors (ICIs) for patients with glioblastoma (GBM) TMZ, temozolomide; CAR T, chimeric antigen receptor T-cell; LITT, laser interstitial thermal therapy; tx, therapy

Meanwhile, the failure of ICIs in some trials of hypermutated and MMR-mutated recurrent high-grade gliomas may reflect the fact that high TMB alone may not be sufficiently predictive of response to ICIs [52]. This finding was corroborated by McGrail and colleagues who demonstrated that high TMB did not predict a positive response to ICI in gliomas [53]. The deleterious impact of temozolomide on T-cell exhaustion could be a contributing force limiting the effectiveness of ICI in the recurrent setting – this needs to be investigated further [49].

Among other potential biomarkers of ICI response, alterations in the mitogen-activated protein kinase (MAPK) pathway, have been associated with improved responses to immunotherapies in other cancers, warranting their further investigation in GBM as well. A secondary analysis of NCT04775485 revealed that responders exhibited an enrichment of MAPK (ERK) pathway alterations, specifically BRAF and PTPN11 mutations [54]. This observation parallels findings in melanoma, where such mutations are associated with improved responses to immunotherapies. Conversely, non-responsive GBM tumors were significantly enriched for PTEN mutations, which are linked to immunosuppressive gene signatures and therapy resistance [55]. Notably, RAS-MAPK pathway blockade, combined with continued ICI therapy, was associated with reboosted anti-tumor immune responses and corresponding radiologic responses [56]. Similarly, in melanoma, initial sensitivity to BRAF inhibitors often gives way to resistance through activation of alternative pathways like MEK or PI3K, and PTPN11 mutations, that further enhance MAPK signaling [57]. Results are awaited from the Alliance A071702 phase II study of ICIs in patients with somatically hypermutated recurrent GBM, a subgroup that may exhibit increased susceptibility to immunotherapy, potentially due to MAPK pathway alterations driving immune responses (NCT04145115). MAPK alterations may influence ICI response, similar to their role in other cancers such as melanoma. ERK1/2 phosphorylation (phospho-ERK), has been recently demonstrated to predict mOS post-ICI therapy in independent cohorts of recurrent GBM patients [58]. Here, patients with high tumor p-ERK had greater tumor-infiltrating myeloid cells and microglia with increased MHC Class II expression [58].

Further research, leveraging both preoperative sequencing of archival tissue and imaging analyses, along with investigation of treated tumor tissue, will provide insights into the exact biological underpinnings of GBM response to ICI and identification of predictive signatures (Fig. 3). Multi-omic analyses have revealed ICI response in GBM patients is associated with distinct immunological attributes, such as T-cell clonal diversity and immune cell tumor infiltration, and genomic signatures, such as PTEN mutations [59]. ICI-non-responder GBM patients are now recognize to have possess primary resistance to ICI, while responder patients show steadily increasing resistance after selection pressure on-therapy [60]. Pre-specified, prospective validation studies are now warranted to translate these novel prognostic signatures to clinical practice (Fig. 4).

Strategies for optimizing ICI trials

Longitudinal tissue sampling

Secondary analyses of ICI “responders” has provided indications that certain tumor subgroups are more likely to have a favorable response to ICI over others (Fig. 3) [61, 62]. Though frequent tissue sampling of GBM tumor in participants within clinical trials can provide critical insight into the molecular evolution within the TME, traditionally this has not been considered safe and feasible. This notion has been challenged recently by some investigators who propose that serial tissue sampling may be justified as it provides critical, actionable insights that outweigh the associated risks [63]. Through real-time analysis of T cell interactions with the tumor, frequent sampling allows for anticipating and countering resistance mechanisms, based on longitudinal molecular evaluation. By ensuring that each tumor tissue sampling event is strategically timed and focused, the risks can be minimized, making the practice safe and feasible, when managed carefully with modern tools.

Role of molecular imaging, radiomics, and radiogenomics

Advances in imaging tools have been crucial in the non-invasive evaluation of GBM and in assessing the response to immunotherapy. The Immunotherapy Response Assessment for Neuro-Oncology (iRANO) criteria, adapted from the immune-related response assessment criteria for solid tumors (iRECIST), provide for a standardized approach to differentiate between true tumor progression and pseudoprogression, which is a common challenge in patients undergoing immunotherapy [64].

Additional advanced imaging techniques are being developed to provide a more granular evaluation of tumor responses to immunotherapy. For instance, Antonios et al. have developed a non-invasive combination imaging technique that differentiates immune-related inflammatory changes from tumor progression in intracranial murine gliomas and GBM patients treated with dendritic cell vaccination and/or PD-1 inhibition [65]. This approach combines MRI and PET imaging with a probe for deoxycytidine kinase (dCK), a protein marker overexpressed in immune cells. Increased dCK activity correlated closely with immune cell infiltration, offering a reliable biomarker for monitoring ICI response. Similarly, PET imaging has been used to preclinically determine the infiltration of CD8+, CD11b+, and CD45 + immune cells in tumors in response to anti-PD-1 therapy [66]. These approaches offer molecular imaging of immunotherapy responses in GBM [67]. However, imaging modalities, including MRI and PET, have limitations. They may struggle to capture dynamic molecular changes and can find it challenging to distinguish between true tumor progression and pseudoprogression [68, 69]. Tools for non-invasive longitudinal assessment of treatment efficacy, such as oncologic molecular imaging, will help tailor therapies to individual patient needs in real-time (Fig. 4).

Radiomics has emerged as a powerful tool for developing new imaging markers by extracting subtle image attributes that capture unique aspects of the TME, reflecting key hallmarks of tumor biology. Similarly, radiogenomics has enabled the identification of statistical correlations between genomic profiles and imaging phenotypes. While radiomic [70] and radiogenomic [71, 72] features have been widely used to correlate with tissue biology and clinical outcomes, particularly in the context of chemotherapy, there have been relatively few studies evaluating their role in predicting responses to immunotherapy in GBM. In a post hoc analysis of a multicenter Phase II study (NCT02336165) of patients with GBM undergoing durvalumab therapy (n = 113) using radiomics, a concordance index of 0.692–0.750 was achieved for OS, and a concordance index of 0.680–0.715 was observed for PFS, for a radiomics-based machine learning model [73]. Similarly, Liu et al. demonstrated, on a validation set of 374 patients, that a radiomics model incorporating 11 features could distinguish tumors with varying ICI prognostic signatures with an accuracy of 94%[74]. A recent systematic review described nine retrospective studies on radiomic and radiogenomic biomarkers for immunotherapy in GBM, highlighting that while these markers hold promise for improving patient selection and monitoring treatment response in immunotherapy, existing studies were constrained by bias and concerns over applicability [75]. Prospective, multi-institutional studies are needed to further develop and validate radiomics-based imaging biomarkers for patient selection and response assessment. Another promising opportunity is to integrate imaging biomarkers with plasma liquid biopsy (LBx), particularly in a longitudinal fashion.

Liquid biopsy

Liquid biopsy (LBx) permits for non-invasive interrogation of either the cerebrospinal fluid (CSF) or the blood, through evaluation of circulating tumor DNA (ctDNA) [76, 77]. However, given the logistical limitations to repeated tissue sampling in brain cancer, LBx offers a non-invasive toolset for patient selection, response prediction and longitudinal evaluation of tumor dynamics (Fig. 4) [78].

Blood-based LBx, has become a critical tool in extracranial solid tumors diagnostics, providing insights into tumor burden and therapeutic response [79]. However, in GBM, the BBB poses a significant challenge by restricting the release of ctDNA into the bloodstream [80]. This results in a very low tumor-associated fraction in plasma cfDNA—often less than 0.1%—making detection exceedingly difficult using conventional methods [81]. One potential breakthrough approach may be transcranial focused ultrasound (FUS), which allows for localized, transient, and non-invasive blood-brain barrier opening (BBBO) [82, 83]. An ongoing multicenter, self-controlled, pivotal trial (NCT05383872) is investigating whether FUS-enabled BBBO leads to ≥ 2-fold rise in blood cfDNA post-procedure [84]. If successful, this could potentially enhance the sensitivity of blood LBx.

Given the historic limitations of conventional blood LBx, ctDNA may be more easily and reliably detected in CSF. Unlike plasma, CSF is in direct contact with the brain and spinal cord, providing a more accurate reflection of the tumor’s genetic landscape. Several studies have demonstrated the utility of CSF in LBx, with emerging research concentrating on the domains of proteomics [85, 86] and DNA methylation [87].

CSF LBx may be used to determine molecular biomarkers of ICI response (Fig. 4). So far, it has been utilized for detecting genetic alterations such as H3K27M mutation, prevalent in diffuse midline gliomas and associated with poor prognosis, and IDH1.R132H mutation [78, 88]. The ability of CSF-based liquid biopsy to detect specific genetic mutations offers a direct method to monitor biomarkers previously identified as predictive of response to ICIs. For instance, the detection of alterations in the MAPK pathway or PTEN mutations through CSF can provide timely insights into the likelihood of a patient’s response to ICIs [89, 90].

Recognizing these possibilities, several proprietary CSF LBx tools are being increasingly investigated such as Caris Assure (Caris Life Sciences, Irving, TX) and Belay Summit Assay (Belay Diagnostics, Chicago, IL). An ongoing multicenter trial (planned N = 400) is evaluating brain biomarker diagnostic performance of Belay Summit assay for patients with diverse CNS tumors [91]. However, CSF-based LBx is not without its logistical challenges and there is a need for standardization of protocols for enhancing their reproducibility and reliability [85].

Ongoing investigations and opportunities on the horizon

The landscape of immunotherapy for gliomas, particularly GBM, is evolving rapidly, with several ongoing clinical trials exploring innovative strategies to enhance the efficacy of ICIs (Table 1). Arrieta et al., have reported use of focused ultrasound-mediated delivery of doxorubicin to the brain combined with PD-1 blockade [92]. This approach temporarily disrupts the BBB, allowing better drug penetration and modulating the immune environment for enhanced responses to PD-1 inhibitors (Fig. 4).

Table 1.

Clinical trials for glioblastoma that are currently ongoing, or recently completed. Original table based on ClinicalTrials.Gov, last updated August 31, 2024

NCT ID	Ph.	Population	Intervention	1° Endpoint(s)	2° Endpoint(s)
NCT02667587 (CheckMate-548)	3	Newly diagnosed MGMT-methylated GBM	Nivo + TMZ + RT post-Sx	mOS, mPFS with BICR	OS at 12 and 24 months, Investigator-Assessed PFS
NCT02336165	2	Newly diagnosed and recurrent GBM cohorts	Durva + RT post-surgery	12-month OS, 6-month PFS and OS	Toxicity, mPFS, mOS, ORR, overall QoL (EORTC-QLQ-C3), brain cancer QoL (EORTC-QLQ-BN-20)
NCT02337491	2	Recurrent GBM	Pembro + Bev	MTD, DLT, 6-month PFS	mPFS, mOS, ORR
NCT03405792 (2-THE-TOP)	2	Newly diagnosed GBM	TTF + TMZ + Pembro post-Sx	mPFS	Toxicity, mOS, Glioma-specific immune changes
NCT02337686	2	Recurrent GBM	Nivo	6-month PFS	mOS, mTTP, ORR, safety
NCT03452579	2	Recurrent GBM, post-failure of prior treatments	Nivo + Bev (Standard vs. low dose)	12-month OS	PFS at 6 months, mOS, ORR, mPFS, DOR, Toxicity
NCT03743662	2	MGMT-methylated Recurrent GBM	RT + Bev + Nivo	mOS	6-month PFS, mPFS, ORR
NCT03661723	2	Recurrent GBM (bevacizumab naive and resistant subcohorts)	Pembro + RT	ORR, OS at 6 and 12 months	Toxicity, DOR, mPFS, mOS, 6-month PFS
NCT04195139 (NUTMEG)	2	Newly diagnosed elderly GBM (aged 65 or above)	RT +TMZ + Nivo	mOS	PFS, Toxicity, overall QoL (EORTC-QLQ-C3 & EuroQOL-5D-5 L), brain cancer QoL (EORTC-QLQ-BN-20), NCF, concordance between RANO & iRANO
NCT03890952	2	Recurrent GBM (salvage re-resection and non-re-resection cohorts)	Nivo + Bev	Whole exome and mRNA sequencing	PFS at 6 months
NCT04817254	2/3	Newly diagnosed GBM or gliosarcoma	Nivolumab + Ipi + TMZ	Correlation between immune response and OS	QoL (MDASI-BT), Other correlations between T-cell responses and disease progression/death
NCT04396860	3	Newly diagnosed MGMT-unmethylated GBM	RT + Nivo + Ipi	mPFS and mOS	OS at 2 years, Toxicity, QoL (MDASI-BT), NCF, PRO-CTCAE
NCT03277638	1/2	Recurrent GBM	Pembro + LITT	Timing of LITT (Ph 1), Response (Ph II)	mPFS, mOS, PFS at 6, 12 and 24 months
NCT04977375	1b/2	Recurrent GBM	Pembro + SRS + Surgery	Toxicity, mOS	mPFS, Immune response
NCT05463848	2	Recurrent GBM	Pembro + Olaparib + TMZ	TIL density (Immune response), 6-month PFS	ORR, mOS, Toxicity, Gene expression profiling score
NCT03347617	2	Newly diagnosed GBM	Ferumoxytol MRI for assessing Pembro response	Diagnostic accuracy	PFS, OS, ORR, DOR, Toxicity
NCT04160494	1	Recurrent GBM	Atezo + D2C7-Immunotoxin using CED	Toxicity	-
NCT06069726 (MO AB)	1/2	Recurrent GBM	Pre-Surgery Atezo	mOS	Toxicity, mPFS
NCT03673787 (IceCAP)	1/2	Multi-cohort trial with GBM cohort	Atezo + Ipatasertib	Toxicity, MTD	Immunological & TME changes
NCT04826393	1/2	Recurrent GBM	Cemiplimab+ASP8374	MTD and RP2D, TIE density	Toxicity, 6-month PFS, 12-month OS
NCT06097975 (NEO-GLITIPNI)	1	Recurrent GBM	Neoadjuvant IV Nivo + Ipi → MSR →intracavitary Nivo + Ipi → Adjuvant Nivo + Ipi	Toxicity	ORR at 4-weeks post neoadjuvant therapy, mPFS, mOS, 6-month OS and PFS
NCT05909618	1/2	Recurrent GBM (cohort 2) and MGMT-unmethylated newly diagnosed GBM (cohort 3)	Anti-P-selectin Crizanlizumab +Nivo	Toxicity, Treatment discontinuation	ORR, mPFS, moS, Overall QoL (EORTC), Disease control rate
NCT04145115	2	Recurrent GBM with high TMB	Nivo + Ipi	Safety, efficacy	OS at 12 months, PFS
NCT04013672	2	Recurrent GBM	Pembro + SurVaxM + Sagramostim + Montanide ISA 51	PFS-6	Toxicity at 12 months
NCT05700955	1	Recurrent GBM	Pembro + TMZ	Toxicity	OS, Overall QoL at 24 months
NCT04479241	2	Recurrent GBM	Pembro + Lerapolturev	ORR, DOR, DRR at 24 months	DCR, DDC, PFS, at 24 months

Abbreviations: Atezo, atezolizumab; Bev, bevacizumab; DOR, duration of response; RT, radiation therapy; TMZ, temozolomide; Nivo, nivolumab; Ipi, Ipilimumab; Pembro, pembrolizumab; ORR, overall objective response rate; TTR, Time to Response; OS, overall survival; mOS, median OS; PFS, progression-free survival; mPFS, median PFS; QoL; quality of life; EORTC-QLQ-C30, European Organisation for Research and Treatment of Cancer, EORTC-QLQ-C30, EORTC Core Quality of Life Questionnaire; RANO, response assessment in neuro-oncology; EORTC-QLQ-BN-20, EORTC Brain Cancer Quality of Life Questionnaire; MDASI-BT, MD Anderson Symptom Inventory for Brain Tumor; NCF, Neurocognitive function; PRO-CTCAE, Patient-Reported Outcomes Version of the Common Terminology Criteria for Adverse Events; RANO, response assessment in neuro-oncology criteria; iRANO, immune-based RANO; Sx, surgery; Ph, Phase; TME, Tumor microenvironment; TIL, Tumor Infiltrating lymphocytes; CED, convection enhanced delivery; TMB, tumor mutational burden

The NCT04485949 trial is evaluating the efficacy of combining nivolumab with hypofractionated SRS in patients with recurrent GBM [22]. This trial is based on the rationale that localized high-dose radiation may increase tumor antigen presentation, thereby improving the effectiveness of ICIs. Early findings suggest that radiation could enhance the immune response elicited by nivolumab [22]. Additionally, durvalumab and atezolizumab are currently being explored for their efficacy in GBM, inspired by their success in other malignancies. An opportunity exists for careful combinatorial approaches using dual ICI approach (Fig. 4).

Advancements in drug delivery systems such as nanoparticles are also gaining attention. STING agonist-loaded, CD47/PD-L1-targeting nanoparticles may potentiate anti-tumor immunity and enhance the effects of radiotherapy in GBM models [93]. This dual-targeting approach not only boosts the immune response against tumor cells but also sensitizes the tumor to radiotherapy, potentially overcoming the intrinsic resistance seen in GBM.

Combining ICIs with other therapeutic modalities, such as oncolytic virotherapy (Fig. 4), also represents a promising strategy. Nassiri et al. reported encouraging results from a phase 1/2 trial combining the oncolytic virus DNX-2401, which selectively replicates within tumor cells to induce oncolysis and stimulate an anti-tumor immune response, with pembrolizumab in patients with recurrent GBM, showing enhanced anti-tumor activity and improved patient outcomes (NCT02798406) [94].

Thes efforts should potentially be grounded in longitudinal monitoring of patient quality of life [95], and careful mitigation of immune-related AEs [96], in order to truly improve patient outcomes. Finally, cost-effectiveness analyses and efforts to develop biosimilars for these novel approaches will help expand the accessibility and availability of these therapies [97].

Conclusions

ICI therapy for CNS solid tumors will require a nuanced and bespoke approach based on an integrated understanding of the active tumor biology and current status of the patient’s immune system. The broad adoption of immunotherapy for GBM will likely include combination approaches that span a spectrum of immunotherapeutic agents, adjunctive drug delivery tools, and companion biomarker-based monitoring strategies. It is also likely that integrated radiomic, tissue and plasma-based analyses (including upfront and serial), will uncover immunotherapeutic response signatures, enabling early and adaptive therapeutic adjustments. Through these efforts and technological advancements, the promise and progress observed to date in the field of cancer immunotherapy may be realized on neuro-oncology.

Abbreviations

APC

Antigen-Presenting Cell

bFGF

Basic Fibroblast Growth Factor

CAR-T

Chimeric Antigen Receptor T Cell

CD

Cluster of Differentiation

CNS

Central Nervous System

CR

Complete Response

ctDNA

Circulating Tumor DNA

CTLA-4

Cytotoxic T-Lymphocyte Associated Protein 4

ECM

Extracellular Matrix

EGFRvIII

Epidermal Growth Factor Receptor Variant III

ERK1/2

Extracellular Signal-Regulated Kinases 1/2

Fc-γR

Fc Gamma ReceptorGBM - Glioblastoma

H3K27M

Histone 3 Lysine 27 to Methionine Mutation

ICI

Imune Checkpoint Inhibitor

IFN-γ

Interferon Gamma

IL

Interleukin

LAG-3

Lymphocyte-Activation Gene 3

MAPK

Mitogen-Activated Protein Kinase

MDSC

Myeloid-Derived Suppressor Cells

MGMT

O6-Methylguanine-DNA Methyltransferase

MSI

Microsatellite Instability

OS

Overall Survival

PD-1

Programmed Cell Death Protein 1

PD-L1

Programmed Death-Ligand 1

PFS

Progression-Free Survival

PGE2

Prostaglandin E2

POLE

Polymerase (DNA) Epsilon, Catalytic Subunit

PR

Partial Response

RCT

Randomized Controlled Trial

SHP-2

Src Homology Region 2 Domain–Containing Phosphatase-2

SOC

Standard of care

SOS

Son of Sevenless

STING

Stimulator of Interferon Genes

TAM

Tumor-Associated Microglia and Macrophages

TGF-β

Transforming Growth Factor Beta

TIM-3

T-Cell Immunoglobulin and Mucin Domain-Containing Protein 3

TMB

Tumor Mutational Burden

TME

Tumor Microenvironment

Treg

Regulatory T Cells

Data availability

No datasets were generated or analysed during the current study.